glebm-nokogiri 1.4.2.1

Sign up to get free protection for your applications and to get access to all the features.
Files changed (277) hide show
  1. data/.autotest +26 -0
  2. data/CHANGELOG.ja.rdoc +411 -0
  3. data/CHANGELOG.rdoc +397 -0
  4. data/Manifest.txt +276 -0
  5. data/README.ja.rdoc +106 -0
  6. data/README.rdoc +132 -0
  7. data/Rakefile +183 -0
  8. data/bin/nokogiri +49 -0
  9. data/deps.rip +5 -0
  10. data/ext/nokogiri/extconf.rb +97 -0
  11. data/ext/nokogiri/html_document.c +154 -0
  12. data/ext/nokogiri/html_document.h +10 -0
  13. data/ext/nokogiri/html_element_description.c +276 -0
  14. data/ext/nokogiri/html_element_description.h +10 -0
  15. data/ext/nokogiri/html_entity_lookup.c +32 -0
  16. data/ext/nokogiri/html_entity_lookup.h +8 -0
  17. data/ext/nokogiri/html_sax_parser_context.c +94 -0
  18. data/ext/nokogiri/html_sax_parser_context.h +11 -0
  19. data/ext/nokogiri/nokogiri.c +95 -0
  20. data/ext/nokogiri/nokogiri.h +153 -0
  21. data/ext/nokogiri/xml_attr.c +94 -0
  22. data/ext/nokogiri/xml_attr.h +9 -0
  23. data/ext/nokogiri/xml_attribute_decl.c +70 -0
  24. data/ext/nokogiri/xml_attribute_decl.h +9 -0
  25. data/ext/nokogiri/xml_cdata.c +56 -0
  26. data/ext/nokogiri/xml_cdata.h +9 -0
  27. data/ext/nokogiri/xml_comment.c +54 -0
  28. data/ext/nokogiri/xml_comment.h +9 -0
  29. data/ext/nokogiri/xml_document.c +464 -0
  30. data/ext/nokogiri/xml_document.h +23 -0
  31. data/ext/nokogiri/xml_document_fragment.c +48 -0
  32. data/ext/nokogiri/xml_document_fragment.h +10 -0
  33. data/ext/nokogiri/xml_dtd.c +202 -0
  34. data/ext/nokogiri/xml_dtd.h +10 -0
  35. data/ext/nokogiri/xml_element_content.c +123 -0
  36. data/ext/nokogiri/xml_element_content.h +10 -0
  37. data/ext/nokogiri/xml_element_decl.c +69 -0
  38. data/ext/nokogiri/xml_element_decl.h +9 -0
  39. data/ext/nokogiri/xml_encoding_handler.c +79 -0
  40. data/ext/nokogiri/xml_encoding_handler.h +8 -0
  41. data/ext/nokogiri/xml_entity_decl.c +110 -0
  42. data/ext/nokogiri/xml_entity_decl.h +10 -0
  43. data/ext/nokogiri/xml_entity_reference.c +52 -0
  44. data/ext/nokogiri/xml_entity_reference.h +9 -0
  45. data/ext/nokogiri/xml_io.c +31 -0
  46. data/ext/nokogiri/xml_io.h +11 -0
  47. data/ext/nokogiri/xml_namespace.c +84 -0
  48. data/ext/nokogiri/xml_namespace.h +13 -0
  49. data/ext/nokogiri/xml_node.c +1347 -0
  50. data/ext/nokogiri/xml_node.h +13 -0
  51. data/ext/nokogiri/xml_node_set.c +418 -0
  52. data/ext/nokogiri/xml_node_set.h +9 -0
  53. data/ext/nokogiri/xml_processing_instruction.c +56 -0
  54. data/ext/nokogiri/xml_processing_instruction.h +9 -0
  55. data/ext/nokogiri/xml_reader.c +665 -0
  56. data/ext/nokogiri/xml_reader.h +10 -0
  57. data/ext/nokogiri/xml_relax_ng.c +168 -0
  58. data/ext/nokogiri/xml_relax_ng.h +9 -0
  59. data/ext/nokogiri/xml_sax_parser.c +286 -0
  60. data/ext/nokogiri/xml_sax_parser.h +39 -0
  61. data/ext/nokogiri/xml_sax_parser_context.c +159 -0
  62. data/ext/nokogiri/xml_sax_parser_context.h +10 -0
  63. data/ext/nokogiri/xml_sax_push_parser.c +115 -0
  64. data/ext/nokogiri/xml_sax_push_parser.h +9 -0
  65. data/ext/nokogiri/xml_schema.c +205 -0
  66. data/ext/nokogiri/xml_schema.h +9 -0
  67. data/ext/nokogiri/xml_syntax_error.c +58 -0
  68. data/ext/nokogiri/xml_syntax_error.h +13 -0
  69. data/ext/nokogiri/xml_text.c +50 -0
  70. data/ext/nokogiri/xml_text.h +9 -0
  71. data/ext/nokogiri/xml_xpath_context.c +276 -0
  72. data/ext/nokogiri/xml_xpath_context.h +9 -0
  73. data/ext/nokogiri/xslt_stylesheet.c +142 -0
  74. data/ext/nokogiri/xslt_stylesheet.h +9 -0
  75. data/lib/nokogiri.rb +133 -0
  76. data/lib/nokogiri/css.rb +25 -0
  77. data/lib/nokogiri/css/generated_parser.rb +669 -0
  78. data/lib/nokogiri/css/generated_tokenizer.rb +145 -0
  79. data/lib/nokogiri/css/node.rb +99 -0
  80. data/lib/nokogiri/css/parser.rb +82 -0
  81. data/lib/nokogiri/css/parser.y +232 -0
  82. data/lib/nokogiri/css/syntax_error.rb +7 -0
  83. data/lib/nokogiri/css/tokenizer.rb +7 -0
  84. data/lib/nokogiri/css/tokenizer.rex +55 -0
  85. data/lib/nokogiri/css/xpath_visitor.rb +169 -0
  86. data/lib/nokogiri/decorators/slop.rb +33 -0
  87. data/lib/nokogiri/ffi/encoding_handler.rb +42 -0
  88. data/lib/nokogiri/ffi/html/document.rb +28 -0
  89. data/lib/nokogiri/ffi/html/element_description.rb +81 -0
  90. data/lib/nokogiri/ffi/html/entity_lookup.rb +16 -0
  91. data/lib/nokogiri/ffi/html/sax/parser_context.rb +38 -0
  92. data/lib/nokogiri/ffi/io_callbacks.rb +42 -0
  93. data/lib/nokogiri/ffi/libxml.rb +386 -0
  94. data/lib/nokogiri/ffi/structs/common_node.rb +38 -0
  95. data/lib/nokogiri/ffi/structs/html_elem_desc.rb +24 -0
  96. data/lib/nokogiri/ffi/structs/html_entity_desc.rb +13 -0
  97. data/lib/nokogiri/ffi/structs/xml_alloc.rb +16 -0
  98. data/lib/nokogiri/ffi/structs/xml_attr.rb +19 -0
  99. data/lib/nokogiri/ffi/structs/xml_attribute.rb +27 -0
  100. data/lib/nokogiri/ffi/structs/xml_buffer.rb +16 -0
  101. data/lib/nokogiri/ffi/structs/xml_char_encoding_handler.rb +11 -0
  102. data/lib/nokogiri/ffi/structs/xml_document.rb +117 -0
  103. data/lib/nokogiri/ffi/structs/xml_dtd.rb +28 -0
  104. data/lib/nokogiri/ffi/structs/xml_element.rb +26 -0
  105. data/lib/nokogiri/ffi/structs/xml_element_content.rb +17 -0
  106. data/lib/nokogiri/ffi/structs/xml_entity.rb +32 -0
  107. data/lib/nokogiri/ffi/structs/xml_enumeration.rb +12 -0
  108. data/lib/nokogiri/ffi/structs/xml_node.rb +28 -0
  109. data/lib/nokogiri/ffi/structs/xml_node_set.rb +53 -0
  110. data/lib/nokogiri/ffi/structs/xml_notation.rb +11 -0
  111. data/lib/nokogiri/ffi/structs/xml_ns.rb +15 -0
  112. data/lib/nokogiri/ffi/structs/xml_parser_context.rb +19 -0
  113. data/lib/nokogiri/ffi/structs/xml_relax_ng.rb +14 -0
  114. data/lib/nokogiri/ffi/structs/xml_sax_handler.rb +51 -0
  115. data/lib/nokogiri/ffi/structs/xml_sax_push_parser_context.rb +124 -0
  116. data/lib/nokogiri/ffi/structs/xml_schema.rb +13 -0
  117. data/lib/nokogiri/ffi/structs/xml_syntax_error.rb +31 -0
  118. data/lib/nokogiri/ffi/structs/xml_text_reader.rb +12 -0
  119. data/lib/nokogiri/ffi/structs/xml_xpath_context.rb +38 -0
  120. data/lib/nokogiri/ffi/structs/xml_xpath_object.rb +35 -0
  121. data/lib/nokogiri/ffi/structs/xml_xpath_parser_context.rb +20 -0
  122. data/lib/nokogiri/ffi/structs/xslt_stylesheet.rb +13 -0
  123. data/lib/nokogiri/ffi/weak_bucket.rb +40 -0
  124. data/lib/nokogiri/ffi/xml/attr.rb +41 -0
  125. data/lib/nokogiri/ffi/xml/attribute_decl.rb +27 -0
  126. data/lib/nokogiri/ffi/xml/cdata.rb +19 -0
  127. data/lib/nokogiri/ffi/xml/comment.rb +18 -0
  128. data/lib/nokogiri/ffi/xml/document.rb +162 -0
  129. data/lib/nokogiri/ffi/xml/document_fragment.rb +21 -0
  130. data/lib/nokogiri/ffi/xml/dtd.rb +67 -0
  131. data/lib/nokogiri/ffi/xml/element_content.rb +43 -0
  132. data/lib/nokogiri/ffi/xml/element_decl.rb +19 -0
  133. data/lib/nokogiri/ffi/xml/entity_decl.rb +36 -0
  134. data/lib/nokogiri/ffi/xml/entity_reference.rb +19 -0
  135. data/lib/nokogiri/ffi/xml/namespace.rb +44 -0
  136. data/lib/nokogiri/ffi/xml/node.rb +556 -0
  137. data/lib/nokogiri/ffi/xml/node_set.rb +149 -0
  138. data/lib/nokogiri/ffi/xml/processing_instruction.rb +20 -0
  139. data/lib/nokogiri/ffi/xml/reader.rb +232 -0
  140. data/lib/nokogiri/ffi/xml/relax_ng.rb +85 -0
  141. data/lib/nokogiri/ffi/xml/sax/parser.rb +135 -0
  142. data/lib/nokogiri/ffi/xml/sax/parser_context.rb +67 -0
  143. data/lib/nokogiri/ffi/xml/sax/push_parser.rb +51 -0
  144. data/lib/nokogiri/ffi/xml/schema.rb +109 -0
  145. data/lib/nokogiri/ffi/xml/syntax_error.rb +98 -0
  146. data/lib/nokogiri/ffi/xml/text.rb +18 -0
  147. data/lib/nokogiri/ffi/xml/xpath.rb +9 -0
  148. data/lib/nokogiri/ffi/xml/xpath_context.rb +148 -0
  149. data/lib/nokogiri/ffi/xslt/stylesheet.rb +53 -0
  150. data/lib/nokogiri/html.rb +35 -0
  151. data/lib/nokogiri/html/builder.rb +35 -0
  152. data/lib/nokogiri/html/document.rb +90 -0
  153. data/lib/nokogiri/html/document_fragment.rb +36 -0
  154. data/lib/nokogiri/html/element_description.rb +23 -0
  155. data/lib/nokogiri/html/entity_lookup.rb +13 -0
  156. data/lib/nokogiri/html/sax/parser.rb +48 -0
  157. data/lib/nokogiri/html/sax/parser_context.rb +16 -0
  158. data/lib/nokogiri/syntax_error.rb +4 -0
  159. data/lib/nokogiri/version.rb +37 -0
  160. data/lib/nokogiri/version_warning.rb +14 -0
  161. data/lib/nokogiri/xml.rb +67 -0
  162. data/lib/nokogiri/xml/attr.rb +14 -0
  163. data/lib/nokogiri/xml/attribute_decl.rb +18 -0
  164. data/lib/nokogiri/xml/builder.rb +418 -0
  165. data/lib/nokogiri/xml/cdata.rb +11 -0
  166. data/lib/nokogiri/xml/character_data.rb +7 -0
  167. data/lib/nokogiri/xml/document.rb +194 -0
  168. data/lib/nokogiri/xml/document_fragment.rb +77 -0
  169. data/lib/nokogiri/xml/dtd.rb +11 -0
  170. data/lib/nokogiri/xml/element_content.rb +36 -0
  171. data/lib/nokogiri/xml/element_decl.rb +13 -0
  172. data/lib/nokogiri/xml/entity_decl.rb +19 -0
  173. data/lib/nokogiri/xml/namespace.rb +13 -0
  174. data/lib/nokogiri/xml/node.rb +793 -0
  175. data/lib/nokogiri/xml/node/save_options.rb +42 -0
  176. data/lib/nokogiri/xml/node_set.rb +325 -0
  177. data/lib/nokogiri/xml/notation.rb +6 -0
  178. data/lib/nokogiri/xml/parse_options.rb +85 -0
  179. data/lib/nokogiri/xml/pp.rb +2 -0
  180. data/lib/nokogiri/xml/pp/character_data.rb +18 -0
  181. data/lib/nokogiri/xml/pp/node.rb +56 -0
  182. data/lib/nokogiri/xml/processing_instruction.rb +8 -0
  183. data/lib/nokogiri/xml/reader.rb +74 -0
  184. data/lib/nokogiri/xml/relax_ng.rb +32 -0
  185. data/lib/nokogiri/xml/sax.rb +4 -0
  186. data/lib/nokogiri/xml/sax/document.rb +160 -0
  187. data/lib/nokogiri/xml/sax/parser.rb +115 -0
  188. data/lib/nokogiri/xml/sax/parser_context.rb +16 -0
  189. data/lib/nokogiri/xml/sax/push_parser.rb +60 -0
  190. data/lib/nokogiri/xml/schema.rb +57 -0
  191. data/lib/nokogiri/xml/syntax_error.rb +47 -0
  192. data/lib/nokogiri/xml/text.rb +9 -0
  193. data/lib/nokogiri/xml/xpath.rb +10 -0
  194. data/lib/nokogiri/xml/xpath/syntax_error.rb +11 -0
  195. data/lib/nokogiri/xml/xpath_context.rb +16 -0
  196. data/lib/nokogiri/xslt.rb +48 -0
  197. data/lib/nokogiri/xslt/stylesheet.rb +25 -0
  198. data/lib/xsd/xmlparser/nokogiri.rb +90 -0
  199. data/tasks/cross_compile.rb +158 -0
  200. data/tasks/test.rb +94 -0
  201. data/test/css/test_nthiness.rb +159 -0
  202. data/test/css/test_parser.rb +282 -0
  203. data/test/css/test_tokenizer.rb +190 -0
  204. data/test/css/test_xpath_visitor.rb +85 -0
  205. data/test/ffi/test_document.rb +35 -0
  206. data/test/files/2ch.html +108 -0
  207. data/test/files/address_book.rlx +12 -0
  208. data/test/files/address_book.xml +10 -0
  209. data/test/files/bar/bar.xsd +4 -0
  210. data/test/files/dont_hurt_em_why.xml +422 -0
  211. data/test/files/exslt.xml +8 -0
  212. data/test/files/exslt.xslt +35 -0
  213. data/test/files/foo/foo.xsd +4 -0
  214. data/test/files/po.xml +32 -0
  215. data/test/files/po.xsd +66 -0
  216. data/test/files/shift_jis.html +10 -0
  217. data/test/files/shift_jis.xml +5 -0
  218. data/test/files/snuggles.xml +3 -0
  219. data/test/files/staff.dtd +10 -0
  220. data/test/files/staff.xml +59 -0
  221. data/test/files/staff.xslt +32 -0
  222. data/test/files/tlm.html +850 -0
  223. data/test/files/valid_bar.xml +2 -0
  224. data/test/helper.rb +169 -0
  225. data/test/html/sax/test_parser.rb +74 -0
  226. data/test/html/sax/test_parser_context.rb +48 -0
  227. data/test/html/test_builder.rb +164 -0
  228. data/test/html/test_document.rb +398 -0
  229. data/test/html/test_document_encoding.rb +77 -0
  230. data/test/html/test_document_fragment.rb +182 -0
  231. data/test/html/test_element_description.rb +98 -0
  232. data/test/html/test_named_characters.rb +14 -0
  233. data/test/html/test_node.rb +181 -0
  234. data/test/html/test_node_encoding.rb +27 -0
  235. data/test/test_convert_xpath.rb +135 -0
  236. data/test/test_css_cache.rb +45 -0
  237. data/test/test_encoding_handler.rb +46 -0
  238. data/test/test_memory_leak.rb +87 -0
  239. data/test/test_nokogiri.rb +138 -0
  240. data/test/test_reader.rb +386 -0
  241. data/test/test_soap4r_sax.rb +52 -0
  242. data/test/test_xslt_transforms.rb +188 -0
  243. data/test/xml/node/test_save_options.rb +20 -0
  244. data/test/xml/node/test_subclass.rb +44 -0
  245. data/test/xml/sax/test_parser.rb +307 -0
  246. data/test/xml/sax/test_parser_context.rb +63 -0
  247. data/test/xml/sax/test_push_parser.rb +139 -0
  248. data/test/xml/test_attr.rb +38 -0
  249. data/test/xml/test_attribute_decl.rb +82 -0
  250. data/test/xml/test_builder.rb +210 -0
  251. data/test/xml/test_cdata.rb +50 -0
  252. data/test/xml/test_comment.rb +29 -0
  253. data/test/xml/test_document.rb +668 -0
  254. data/test/xml/test_document_encoding.rb +26 -0
  255. data/test/xml/test_document_fragment.rb +180 -0
  256. data/test/xml/test_dtd.rb +82 -0
  257. data/test/xml/test_dtd_encoding.rb +33 -0
  258. data/test/xml/test_element_content.rb +56 -0
  259. data/test/xml/test_element_decl.rb +73 -0
  260. data/test/xml/test_entity_decl.rb +120 -0
  261. data/test/xml/test_entity_reference.rb +21 -0
  262. data/test/xml/test_namespace.rb +68 -0
  263. data/test/xml/test_node.rb +865 -0
  264. data/test/xml/test_node_attributes.rb +34 -0
  265. data/test/xml/test_node_encoding.rb +107 -0
  266. data/test/xml/test_node_reparenting.rb +293 -0
  267. data/test/xml/test_node_set.rb +649 -0
  268. data/test/xml/test_parse_options.rb +52 -0
  269. data/test/xml/test_processing_instruction.rb +30 -0
  270. data/test/xml/test_reader_encoding.rb +126 -0
  271. data/test/xml/test_relax_ng.rb +60 -0
  272. data/test/xml/test_schema.rb +89 -0
  273. data/test/xml/test_syntax_error.rb +12 -0
  274. data/test/xml/test_text.rb +38 -0
  275. data/test/xml/test_unparented_node.rb +381 -0
  276. data/test/xml/test_xpath.rb +138 -0
  277. metadata +533 -0
@@ -0,0 +1,27 @@
1
+ # -*- coding: utf-8 -*-
2
+ require "helper"
3
+
4
+ module Nokogiri
5
+ module HTML
6
+ if RUBY_VERSION =~ /^1\.9/
7
+ class TestNodeEncoding < Nokogiri::TestCase
8
+ def test_inner_html
9
+ doc = Nokogiri::HTML File.open(SHIFT_JIS_HTML, 'rb')
10
+
11
+ hello = "こんにちは"
12
+
13
+ contents = doc.at('h2').inner_html
14
+ assert_equal doc.encoding, contents.encoding.name
15
+ assert_match hello.encode('Shift_JIS'), contents
16
+
17
+ contents = doc.at('h2').inner_html(:encoding => 'UTF-8')
18
+ assert_match hello, contents
19
+
20
+ doc.encoding = 'UTF-8'
21
+ contents = doc.at('h2').inner_html
22
+ assert_match hello, contents
23
+ end
24
+ end
25
+ end
26
+ end
27
+ end
@@ -0,0 +1,135 @@
1
+ require "helper"
2
+
3
+ class TestConvertXPath < Nokogiri::TestCase
4
+
5
+ def setup
6
+ super
7
+ @N = Nokogiri(File.read(HTML_FILE))
8
+ end
9
+
10
+ def assert_syntactical_equivalence(hpath, xpath, match, &blk)
11
+ blk ||= lambda {|j| j.first}
12
+ assert_equal match, blk.call(@N.search(xpath)), "xpath result did not match"
13
+ end
14
+
15
+ def test_child_tag
16
+ assert_syntactical_equivalence("h1[a]", ".//h1[child::a]", "Tender Lovemaking") do |j|
17
+ j.inner_text
18
+ end
19
+ end
20
+
21
+ def test_child_tag_equals
22
+ assert_syntactical_equivalence("h1[a='Tender Lovemaking']", ".//h1[child::a = 'Tender Lovemaking']", "Tender Lovemaking") do |j|
23
+ j.inner_text
24
+ end
25
+ end
26
+
27
+ def test_filter_contains
28
+ assert_syntactical_equivalence("title:contains('Tender')", ".//title[contains(., 'Tender')]",
29
+ "Tender Lovemaking ") do |j|
30
+ j.inner_text
31
+ end
32
+ end
33
+
34
+ def test_filter_comment
35
+ assert_syntactical_equivalence("div comment()[2]", ".//div//comment()[position() = 2]", "<!-- end of header -->") do |j|
36
+ j.first.to_s
37
+ end
38
+ end
39
+
40
+ def test_filter_text
41
+ assert_syntactical_equivalence("a[text()]", ".//a[normalize-space(child::text())]", "<a href=\"http://tenderlovemaking.com\">Tender Lovemaking</a>") do |j|
42
+ j.first.to_s
43
+ end
44
+ assert_syntactical_equivalence("a[text()='Tender Lovemaking']", ".//a[normalize-space(child::text()) = 'Tender Lovemaking']", "<a href=\"http://tenderlovemaking.com\">Tender Lovemaking</a>") do |j|
45
+ j.first.to_s
46
+ end
47
+ assert_syntactical_equivalence("a/text()", ".//a/child::text()", "Tender Lovemaking") do |j|
48
+ j.first.to_s
49
+ end
50
+ assert_syntactical_equivalence("h2//a[text()!='Back Home!']", ".//h2//a[normalize-space(child::text()) != 'Back Home!']", "Meow meow meow meow meow") do |j|
51
+ j.first.inner_text
52
+ end
53
+ end
54
+
55
+ def test_filter_by_attr
56
+ assert_syntactical_equivalence("a[@href='http://blog.geminigeek.com/wordpress-theme']",
57
+ ".//a[@href = 'http://blog.geminigeek.com/wordpress-theme']",
58
+ "http://blog.geminigeek.com/wordpress-theme") do |j|
59
+ j.first["href"]
60
+ end
61
+ end
62
+
63
+ def test_css_id
64
+ assert_syntactical_equivalence("#linkcat-7", ".//*[@id = 'linkcat-7']", "linkcat-7") do |j|
65
+ j.first["id"]
66
+ end
67
+ assert_syntactical_equivalence("li#linkcat-7", ".//li[@id = 'linkcat-7']", "linkcat-7") do |j|
68
+ j.first["id"]
69
+ end
70
+ end
71
+
72
+ def test_css_class
73
+ assert_syntactical_equivalence(".cat-item-15", ".//*[contains(concat(' ', @class, ' '), ' cat-item-15 ')]",
74
+ "cat-item cat-item-15") do |j|
75
+ j.first["class"]
76
+ end
77
+ assert_syntactical_equivalence("li.cat-item-15", ".//li[contains(concat(' ', @class, ' '), ' cat-item-15 ')]",
78
+ "cat-item cat-item-15") do |j|
79
+ j.first["class"]
80
+ end
81
+ end
82
+
83
+ def test_css_tags
84
+ assert_syntactical_equivalence("div li a", ".//div//li//a", "http://brobinius.org/") do |j|
85
+ j.first.inner_text
86
+ end
87
+ assert_syntactical_equivalence("div li > a", ".//div//li/a", "http://brobinius.org/") do |j|
88
+ j.first.inner_text
89
+ end
90
+ assert_syntactical_equivalence("h1 ~ small", ".//small[preceding-sibling::h1]", "The act of making love, tenderly.") do |j|
91
+ j.first.inner_text
92
+ end
93
+ assert_syntactical_equivalence("h1 ~ small", ".//small[preceding-sibling::h1]", "The act of making love, tenderly.") do |j|
94
+ j.first.inner_text
95
+ end
96
+ end
97
+
98
+ def test_positional
99
+ assert_syntactical_equivalence("div/div:first()", ".//div/div[position() = 1]", "\r\nTender Lovemaking\r\nThe act of making love, tenderly.\r\n".gsub(/[\r\n]/, '')) do |j|
100
+ j.first.inner_text.gsub(/[\r\n]/, '')
101
+ end
102
+ assert_syntactical_equivalence("div/div:first", ".//div/div[position() = 1]", "\r\nTender Lovemaking\r\nThe act of making love, tenderly.\r\n".gsub(/[\r\n]/, '')) do |j|
103
+ j.first.inner_text.gsub(/[\r\n]/, '')
104
+ end
105
+ assert_syntactical_equivalence("div//a:last()", ".//div//a[position() = last()]", "Wordpress") do |j|
106
+ j.last.inner_text
107
+ end
108
+ assert_syntactical_equivalence("div//a:last", ".//div//a[position() = last()]", "Wordpress") do |j|
109
+ j.last.inner_text
110
+ end
111
+ end
112
+
113
+ def test_multiple_filters
114
+ assert_syntactical_equivalence("a[@rel='bookmark'][1]", ".//a[@rel = 'bookmark' and position() = 1]", "Back Home!") do |j|
115
+ j.first.inner_text
116
+ end
117
+ end
118
+
119
+ # TODO:
120
+ # doc/'title ~ link' -> links that are siblings of title
121
+ # doc/'p[@class~="final"]' -> class includes string (whitespacy)
122
+ # doc/'p[text()*="final"]' -> class includes string (index) (broken: always returns true?)
123
+ # doc/'p[text()$="final"]' -> /final$/
124
+ # doc/'p[text()|="final"]' -> /^final$/
125
+ # doc/'p[text()^="final"]' -> string starts with 'final
126
+ # nth_first
127
+ # nth_last
128
+ # even
129
+ # odd
130
+ # first-child, nth-child, last-child, nth-last-child, nth-last-of-type
131
+ # only-of-type, only-child
132
+ # parent
133
+ # empty
134
+ # root
135
+ end
@@ -0,0 +1,45 @@
1
+ require "helper"
2
+
3
+ class TestCssCache < Nokogiri::TestCase
4
+
5
+ def setup
6
+ super
7
+ @css = "a1 > b2 > c3"
8
+ @parse_result = Nokogiri::CSS.parse(@css)
9
+ @to_xpath_result = @parse_result.map {|ast| ast.to_xpath}
10
+ Nokogiri::CSS::Parser.class_eval do
11
+ class << @cache
12
+ alias :old_bracket :[]
13
+ attr_reader :count
14
+ def [](key)
15
+ @count ||= 0
16
+ @count += 1
17
+ old_bracket(key)
18
+ end
19
+ end
20
+ end
21
+ assert Nokogiri::CSS::Parser.cache_on?
22
+ end
23
+
24
+ def teardown
25
+ Nokogiri::CSS::Parser.clear_cache
26
+ Nokogiri::CSS::Parser.set_cache true
27
+ end
28
+
29
+ [ false, true ].each do |cache_setting|
30
+ define_method "test_css_cache_#{cache_setting ? "true" : "false"}" do
31
+ times = cache_setting ? 4 : nil
32
+
33
+ Nokogiri::CSS::Parser.set_cache cache_setting
34
+
35
+ Nokogiri::CSS.xpath_for(@css)
36
+ Nokogiri::CSS.xpath_for(@css)
37
+ Nokogiri::CSS::Parser.new.xpath_for(@css)
38
+ Nokogiri::CSS::Parser.new.xpath_for(@css)
39
+
40
+ assert_equal(times, Nokogiri::CSS::Parser.class_eval { @cache.count })
41
+ end
42
+ end
43
+
44
+
45
+ end
@@ -0,0 +1,46 @@
1
+ # -*- coding: utf-8 -*-
2
+
3
+ require "helper"
4
+
5
+ class TestEncodingHandler < Nokogiri::TestCase
6
+ def teardown
7
+ Nokogiri::EncodingHandler.clear_aliases!
8
+ end
9
+
10
+ def test_get
11
+ assert_not_nil Nokogiri::EncodingHandler['UTF-8']
12
+ assert_nil Nokogiri::EncodingHandler['alsdkjfhaldskjfh']
13
+ end
14
+
15
+ def test_name
16
+ eh = Nokogiri::EncodingHandler['UTF-8']
17
+ assert_equal "UTF-8", eh.name
18
+ end
19
+
20
+ def test_alias
21
+ Nokogiri::EncodingHandler.alias('UTF-8', 'UTF-18')
22
+ assert_equal 'UTF-8', Nokogiri::EncodingHandler['UTF-18'].name
23
+ end
24
+
25
+ def test_cleanup_aliases
26
+ assert_nil Nokogiri::EncodingHandler['UTF-9']
27
+ Nokogiri::EncodingHandler.alias('UTF-8', 'UTF-9')
28
+ assert_not_nil Nokogiri::EncodingHandler['UTF-9']
29
+
30
+ Nokogiri::EncodingHandler.clear_aliases!
31
+ assert_nil Nokogiri::EncodingHandler['UTF-9']
32
+ end
33
+
34
+ def test_delete
35
+ assert_nil Nokogiri::EncodingHandler['UTF-9']
36
+ Nokogiri::EncodingHandler.alias('UTF-8', 'UTF-9')
37
+ assert_not_nil Nokogiri::EncodingHandler['UTF-9']
38
+
39
+ Nokogiri::EncodingHandler.delete 'UTF-9'
40
+ assert_nil Nokogiri::EncodingHandler['UTF-9']
41
+ end
42
+
43
+ def test_delete_non_existent
44
+ assert_nil Nokogiri::EncodingHandler.delete('UTF-9')
45
+ end
46
+ end
@@ -0,0 +1,87 @@
1
+ require "helper"
2
+
3
+ class TestMemoryLeak < Nokogiri::TestCase
4
+
5
+ if ENV['NOKOGIRI_GC'] # turning these off by default for now
6
+
7
+ def test_dont_hurt_em_why
8
+ content = File.open("#{File.dirname(__FILE__)}/files/dont_hurt_em_why.xml").read
9
+ ndoc = Nokogiri::XML(content)
10
+ 2.times do
11
+ info = ndoc.search('status text').first.inner_text
12
+ url = ndoc.search('user name').first.inner_text
13
+ GC.start
14
+ end
15
+ end
16
+
17
+ def test_for_memory_leak
18
+ begin
19
+ # we don't use Dike in any tests, but requiring it has side effects
20
+ # that can create memory leaks, and that's what we're testing for.
21
+ require 'rubygems'
22
+ require 'dike' # do not remove!
23
+
24
+ count_start = count_object_space_documents
25
+ xml_data = <<-EOS
26
+ <test>
27
+ <items>
28
+ <item>abc</item>
29
+ <item>1234</item>
30
+ <item>Zzz</item>
31
+ <items>
32
+ </test>
33
+ EOS
34
+ 20.times do
35
+ doc = Nokogiri::XML(xml_data)
36
+ doc.xpath("//item")
37
+ end
38
+ 2.times { GC.start }
39
+ count_end = count_object_space_documents
40
+ assert((count_end - count_start) <= 2, "memory leak detected")
41
+ rescue LoadError
42
+ puts "\ndike is not installed, skipping memory leak test"
43
+ end
44
+ end
45
+
46
+ if Nokogiri.ffi?
47
+ [ ['Node', 'p', nil],
48
+ ['CDATA', nil, 'content'],
49
+ ['Comment', nil, 'content'],
50
+ ['DocumentFragment', nil],
51
+ ['EntityReference', nil, 'p'],
52
+ ['ProcessingInstruction', nil, 'p', 'content'] ].each do |klass, *args|
53
+
54
+ define_method "test_for_leaked_#{klass}_nodes" do
55
+ Nokogiri::LibXML.expects(:xmlAddChild).at_least(1) # more than once shows we're GCing properly
56
+ 10.times {
57
+ xml = Nokogiri::XML("<root></root>")
58
+ 2.times { Nokogiri::XML.const_get(klass).new(*(args.collect{|arg| arg || xml})) }
59
+ GC.start
60
+ }
61
+ GC.start
62
+ end
63
+
64
+ end
65
+
66
+ def test_for_leaked_attr_nodes
67
+ Nokogiri::LibXML.expects(:xmlFreePropList).at_least(1) # more than once shows we're GCing properly
68
+ 10.times {
69
+ xml = Nokogiri::XML("<root></root>")
70
+ 2.times { Nokogiri::XML::Attr.new(xml, "p") }
71
+ GC.start
72
+ }
73
+ GC.start
74
+ end
75
+
76
+ end # if ffi
77
+
78
+ end # if NOKOGIRI_GC
79
+
80
+ private
81
+
82
+ def count_object_space_documents
83
+ count = 0
84
+ ObjectSpace.each_object {|j| count += 1 if j.is_a?(Nokogiri::XML::Document) }
85
+ count
86
+ end
87
+ end
@@ -0,0 +1,138 @@
1
+ require "helper"
2
+
3
+ class TestNokogiri < Nokogiri::TestCase
4
+ def test_versions
5
+ version_match = /\d+\.\d+\.\d+/
6
+ assert_match version_match, Nokogiri::VERSION
7
+ assert_match version_match, Nokogiri::LIBXML_VERSION
8
+
9
+ if defined?(FFI) && defined?(Nokogiri::LibXML)
10
+ assert_equal 'ffi', Nokogiri::VERSION_INFO['libxml']['binding']
11
+ if RUBY_PLATFORM =~ /java/
12
+ assert_equal 'jruby', Nokogiri::VERSION_INFO['libxml']['platform']
13
+ else
14
+ assert_equal 'ruby', Nokogiri::VERSION_INFO['libxml']['platform']
15
+ end
16
+ assert Nokogiri.ffi?
17
+ else
18
+ assert_equal 'extension', Nokogiri::VERSION_INFO['libxml']['binding']
19
+
20
+ assert_match version_match, Nokogiri::VERSION_INFO['libxml']['compiled']
21
+ assert_equal Nokogiri::LIBXML_VERSION, Nokogiri::VERSION_INFO['libxml']['compiled']
22
+ assert ! Nokogiri.ffi?
23
+ end
24
+
25
+ assert_match version_match, Nokogiri::VERSION_INFO['libxml']['loaded']
26
+ Nokogiri::LIBXML_PARSER_VERSION =~ /(\d)(\d{2})(\d{2})/
27
+ major = $1.to_i
28
+ minor = $2.to_i
29
+ bug = $3.to_i
30
+ assert_equal "#{major}.#{minor}.#{bug}", Nokogiri::VERSION_INFO['libxml']['loaded']
31
+ end
32
+
33
+ def test_libxml_iconv
34
+ assert Nokogiri.const_defined?(:LIBXML_ICONV_ENABLED)
35
+ end
36
+
37
+ def test_parse_with_io
38
+ doc = Nokogiri.parse(
39
+ StringIO.new("<html><head><title></title><body></body></html>")
40
+ )
41
+ assert_instance_of Nokogiri::HTML::Document, doc
42
+ end
43
+
44
+ def test_xml?
45
+ doc = Nokogiri.parse(File.read(XML_FILE))
46
+ assert doc.xml?
47
+ assert !doc.html?
48
+ end
49
+
50
+ def test_html?
51
+ doc = Nokogiri.parse(File.read(HTML_FILE))
52
+ assert !doc.xml?
53
+ assert doc.html?
54
+ end
55
+
56
+ def test_nokogiri_method_with_html
57
+ doc1 = Nokogiri(File.read(HTML_FILE))
58
+ doc2 = Nokogiri.parse(File.read(HTML_FILE))
59
+ assert_equal doc1.serialize, doc2.serialize
60
+ end
61
+
62
+ def test_nokogiri_method_with_block
63
+ doc = Nokogiri { b "bold tag" }
64
+ assert_equal('<b>bold tag</b>', doc.to_html.chomp)
65
+ end
66
+
67
+ def test_make_with_html
68
+ doc = Nokogiri.make("<b>bold tag</b>")
69
+ assert_equal('<b>bold tag</b>', doc.to_html.chomp)
70
+ end
71
+
72
+ def test_make_with_block
73
+ doc = Nokogiri.make { b "bold tag" }
74
+ assert_equal('<b>bold tag</b>', doc.to_html.chomp)
75
+ end
76
+
77
+ SLOP_HTML = <<-END
78
+ <html>
79
+ <body>
80
+ <ul>
81
+ <li class='red'>one</li>
82
+ <li class='blue'>two</li>
83
+ </ul>
84
+ <div>
85
+ one
86
+ <div>div two</div>
87
+ </div>
88
+ </body>
89
+ </html>
90
+ END
91
+
92
+ def test_slop_css
93
+ doc = Nokogiri::Slop(<<-eohtml)
94
+ <html>
95
+ <body>
96
+ <div>
97
+ one
98
+ <div class='foo'>
99
+ div two
100
+ <div class='foo'>
101
+ div three
102
+ </div>
103
+ </div>
104
+ </div>
105
+ </body>
106
+ </html>
107
+ eohtml
108
+ assert_equal "div", doc.html.body.div.div('.foo').name
109
+ end
110
+
111
+ def test_slop
112
+ doc = Nokogiri::Slop(SLOP_HTML)
113
+
114
+ assert_equal "one", doc.html.body.ul.li.first.text
115
+ assert_equal "two", doc.html.body.ul.li(".blue").text
116
+ assert_equal "div two", doc.html.body.div.div.text
117
+
118
+ assert_equal "two", doc.html.body.ul.li(:css => ".blue").text
119
+
120
+ assert_equal "two", doc.html.body.ul.li(:xpath => "position()=2").text
121
+ assert_equal "one", doc.html.body.ul.li(:xpath => ["contains(text(),'o')"]).first.text
122
+ assert_equal "two", doc.html.body.ul.li(:xpath => ["contains(text(),'o')","contains(text(),'t')"]).text
123
+
124
+ assert_raise(NoMethodError) { doc.nonexistent }
125
+ end
126
+
127
+ def test_slop_decorator
128
+ doc = Nokogiri(SLOP_HTML)
129
+ assert !doc.decorators(Nokogiri::XML::Node).include?(Nokogiri::Decorators::Slop)
130
+
131
+ doc.slop!
132
+ assert doc.decorators(Nokogiri::XML::Node).include?(Nokogiri::Decorators::Slop)
133
+
134
+ doc.slop!
135
+ assert_equal 1, doc.decorators(Nokogiri::XML::Node).select { |d| d == Nokogiri::Decorators::Slop }.size
136
+ end
137
+
138
+ end